Abstract Wil van der Aalst
Process mining provides a bridge between process science and data science and process mining tools are rapidly being adopted as a way to analyze event data in various application domains. Process discovery is the first, and probably most challenging, step in any process mining project. Process discovery techniques tend to return models that are either formal (precisely describing the possible behaviors) or informal (merely a “picture” not allowing for any form of formal reasoning).
Models having formal semantics are able to classify traces, i.e., sequences of events, as fitting or non-fitting. Informal models remain deliberately vague on the set of possible traces. Most process miming approaches described in literature produce formal models. This is in stark contrast with the over 25 available commercial process mining tools that only discover informal models. There are two main reasons vendors use informal models: scalability and simplicity. This talk proposes to combine the best of both worlds: discovering hybrid process models that have formal and informal elements. Initial results demonstrate the advantages of remaining “vague” when there is not enough evidence in the event data.
Moreover, the approach is scalable and be used for larger data sets (exploiting distributed infrastructures). The expectation is that the next generation of commercial process mining tools will adopt this approach, thus combining the best of both worlds.